oqgraph-dev team mailing list archive

[Heinz Wiesinger <HMWiesinger@xxxxxxxxxxx>]

Hi Andrew,

Filed https://mariadb.atlassian.net/browse/MDEV-9450

Thanks again!

Grs,
Heinz

On Friday 22 January 2016 21:40:06 Andrew McDonnell wrote:
> Hi Heinz
> 
> opening a new bug with the stack trace is the best thing to do, please
> 
> HTH
> --Andrew
> 
> On 22/01/16 21:33, Heinz Wiesinger wrote:
> > Hi Andrew,
> > 
> > I just saw you opened pull requests to add the fix we discussed. Thanks a
> > lot for that! :)
> > 
> > I however also noticed it doesn't seem to touch the issue I mentioned
> > below
> > (crash on shutdown). I should've probably mentioned that in the Jira issue
> > as well. Do you want me to open a separate issue for that crash?
> > 
> > Grs,
> > Heinz
> > 
> > On Monday 30 November 2015 14:08:15 Heinz Wiesinger wrote:
> >> Hi Andrew,
> >> 
> >> After a bit more testing I found still one more minor issue. Query
> >> execution still works fine now, but when shutting down mariadb I get a
> >> different crash (see log attached). Let me know if you need more info.
> >> 
> >> Grs,
> >> Heinz
> >> 
> >> On Monday 30 November 2015 13:32:49 Heinz Wiesinger wrote:
> >>> Hi Andrew,
> >>> 
> >>> Sorry for taking my time with this. You surprised me a bit with
> >>> delivering
> >>> a patch so quickly ;)
> >>> 
> >>> It looks like your hunch is correct. After applying the patch my crashes
> >>> are gone.
> >>> 
> >>> Thanks for looking into this and I'm looking forward to seeing the pull
> >>> request :)
> >>> 
> >>> Grs,
> >>> Heinz
> >>> 
> >>> (Sorry Andrew for getting this twice. Forgot to include the list in the
> >>> previous mail)
> >>> 
> >>> On Monday 16 November 2015 23:11:18 you wrote:
> >>>> Hi Heinz
> >>>> 
> >>>> I'm going to take a punt and say that I missed this applying the fix
> >>>> for
> >>>> MDEV-6282 in all cases
> >>>> 
> >>>> If you can manually apply the following patch and see if it fixes your
> >>>> problem?
> >>>> 
> >>>> HTH
> >>>> --Andrew
> >>>> 
> >>>> 
> >>>> 
> >>>> diff --git a/storage/oqgraph/ha_oqgraph.cc
> >>>> b/storage/oqgraph/ha_oqgraph.cc
> >>>> index 89bc912..b4cf03c 100644
> >>>> --- a/storage/oqgraph/ha_oqgraph.cc
> >>>> +++ b/storage/oqgraph/ha_oqgraph.cc
> >>>> @@ -1099,6 +1099,10 @@ int ha_oqgraph::info(uint flag)
> >>>> 
> >>>>  int ha_oqgraph::extra(enum ha_extra_function operation)
> >>>>  {
> >>>> 
> >>>> +  if (graph->get_thd() != current_thd) {
> >>>> +    DBUG_PRINT( "oq-debug", ("rnd_pos g->table->in_use: 0x%lx <--
> >>>> current_thd 0x%lx", (long) graph->get_thd(), (long) current_thd));
> >>>> +    graph->set_thd(current_thd);
> >>>> +  }
> >>>> 
> >>>>    return edges->file->extra(operation);
> >>>>  
> >>>>  }
> >>>> 
> >>>> On 16/11/15 19:17, Heinz Wiesinger wrote:
> >>>>> Hey everyone,
> >>>>> 
> >>>>> We've been playing around a bit more with OQGraphv3 and when trying to
> >>>>> put
> >>>>> it on one of our production setups we found it crashing in situations
> >>>>> that didn't happen on our testing setup. After a bit of back and forth
> >>>>> I
> >>>>> managed to reproduce it pretty reliable and added info to
> >>>>> https://mariadb.atlassian.net/browse/MDEV-8206
> >>>>> 
> >>>>> From what I see it seems that the crash only happens when the graph is
> >>>>> above a certain size. But with our code, we'd hit that limit pretty
> >>>>> easily.
> >>>>> 
> >>>>> Could someone maybe take a look at that please? :)
> >>>>> 
> >>>>> Grs,
> >>>>> Heinz

Attachment: signature.asc
Description: This is a digitally signed message part.